On the Origin of fMRI Species

Abstract

It has been pointed out by the JMRI editor, Mark E. Schweitzer and others that the nomenclature of MRI and, in particular, functional MRI (fMRI) can appear imprecise. While those working within their specialty appreciate the nuance of term, the term meaning can blur with the professional distance from the specialty. This problem has no clear solution other than educating those outside the field and fostering an implicit respect and acceptance that in any mature field, terms do have specific meaning and nuance. Terms in science naturally evolve to efficiently codify and communicate precise information. Terms become more precise over time in any specialty.

It has been pointed out that the term “fMRI” is often confused with “resting state fMRI” or “BOLD.” This could be because fMRI is a field that has grown in awareness among the non-experts perhaps more than other fields due to its popular appeal. We nevertheless found this confusion concerning and needing of immediate clarification.

In fMRI, blood oxygenation dependent (BOLD) has never been used interchangeably with fMRI. BOLD is an endogenous contrast arising from the unique property of blood to change magnetic susceptibility depending on its oxygenation, while fMRI is a general term for methods that use BOLD as one of several functional contrasts. BOLD was coined by Ogawa et al in his 1991 PNAS paper¹ that predates the first successful brain activation experiment. It was not until 1993 when the term fMRI was first used to define the category of approaches that involve the collection of time series of data where neuronal activity induced MR signal changes were observed as they evolved over time.² There was no ambiguity. Functional MRI was used differently in a review article that came out in Nature in 1990³ before fMRI as we know it was discovered. The 1990 review showed the various types of “functional MRI,” including diffusion imaging—pioneered by the co-author on that paper and this editorial, Denis Le Bihan, but was clearly not the same use of the term as it was used since 1993. In the 1990 review, time series collection of images showing functional signal changes was not involved. Should the authors of the 1993 paper have thought twice before adopting that term to time series imaging of contrast changes related to brain function, and could they have come up with a better term to differentiate it from the 1990 article? Perhaps. However, everyone in the field understood precisely what was meant and the term was embraced. At the time this term was introduced, the collection of methods that included exogenous contrast agent based blood volume change imaging⁴ arterial spin-labeling (ASL) based perfusion,⁵ and of course, BOLD contrast^6-8 were well-understood to be in the domain of fMRI approaches. It was also shown later that time series collection of diffusion MRI⁹ and noninvasive blood volume contrast,¹⁰ called vascular space occupancy (VASO), were also in the domain of fMRI. These are functional contrasts that change over time while the brain is changing its level of activity. This is fMRI as we all understand it.

Other acronyms were tossed into the arena in the early days. Robert Turner suggested MRFN (Magnetic Resonance Functional Neuroimaging—or “more fun”), which would have been delightful if it had caught on, but it did not. No one person or even group of people decided on this. The term fMRI may have started with one paper, but it emerged organically and was forged by repeated use. A term, as precise and unique as one tries to make it, ends up being simply what the field comes to agree upon.

The taxonomy of functional MRI terms includes contrasts and paradigms—independent of each other. The activation paradigms include event-related fMRI, block-design fMRI, fMRI adaptation, as well as resting state fMRI. The logic behind the naming of resting state fMRI was that the subject was “at rest” in the scanner, performing no overt or instructed covert task. It is never thought that “resting state activity” is an oxymoron. Brains are spontaneously active when in a “resting-state.” “Task” fMRI and “resting-state” fMRI are both about imaging the functional status of the brain (hence “f” MRI…) over time.

Resting-state fMRI, relying on measures temporal correlation in signal between spontaneously active brain areas, has shown profound utility; however, with the common practice of interchanging the term “correlation” with “connection” the field has gone a bit astray as it has erroneously settled on the term “functional connectivity fMRI” (fcMRI). Temporal correlation does not imply connection—which is an ambiguous term in itself. It is hoped that those who work in the field already know this; however, the term “connectivity” is ambiguous and indeed a source of confusion that will perhaps be rectified as the science advances.

Whether terms catch on or not depends on the dynamically changing consensus of the field. Most terms in a mature field have precise meanings, but as we see with the term fcMRI, there can be exceptions.